Research On Chinese Chess Robots' Binocular Vision Positioning And Kinematic Of Manipulator

With the advent of the artificial intelligence boom and the pursuit of further liberation of productivity and higher quality of life,intelligent robots have developed rapidly.Chinese chess robot belongs to intelligent entertainment robot and is a comprehensive artificial intelligence system.Chinese chess is profound and profound,and it has a long history.It is of far-reaching significance for Chinese chess robots that inherit traditional culture and integrate new technology.This paper focuses on the Chinese chess robot,focusing on its binocular vision positioning and kinematic of manipulator.The ZED binocular camera is selected as the experimental hardware of binocular vision system,and the imaging geometry model of binocular camera is introduced,including linear and nonlinear models.Then,the three methods of camera calibration are compared and the traditional calibration method is used to calibrate the ZED camera for binocular vision.After obtaining the internal and external parameters of the camera,the calibration results are verified,and the camera image is corrected for distortion according to the distortion model.After calibrating the camera and obtaining the connection between the image and the real world,the empty chessboard image taken at the angle of the player is identified to locate the chessboard point.The empty chessboard image is pre-processed and geometrically corrected to obtain a top view of the chessboard,and a chessboard pointdetection algorithm based on the “center point of Jiugongge” is proposed.After obtaining all the chessboard points,calculate the chessboard points in the original image of the camera through reverse mapping,and calculate the world coordinates of chessboard points by the camera's internal and external parameters and triangulation method,the standard chessboard plane is fitted,and the chessboard points are relocated.After obtaining the coordinates of the chessboard point,the chessboard and chess image photographed at the angle of the player is identified to locate the chess.Firstly,through the same method as the empty chessboard image processing,the top view image of the chessboard and the chess piece is obtained through the plane projection transformation matrix.Then the chess piece is detected by the Hough circle detection method,and the center coordinates of the chess piece are obtained.With comparing with the pixel coordinates of the chessboard point,and determing the chess color according to the hue value of neighborhood,the chess piece matrix is generated.At the same time,all the chess pieces are segmented according to the size of the piece.The world physical coordinates of the pieces are calculated by the camera's internal and external parameters and the triangulation method,and the error analysis is performed.After obtaining the coordinates of the chessboard points and the pieces,the binocular vision system provides position information for the robotic execution system.First,an end effector for grasping chess pieces is designed for the human-like manipulator of four rotation degree of freedom.Then the manipulator joint coordinate system is established according to the D-H method,and the inverse kinematics of the manipulator is calculated.Finally,the robot movements between the chessboard points are simulated by Matlab,which proves that the robot can move without dead point on the chessboard.